NAOJ GW Elog Logbook 3.2
We have verified that the spikes oberved in the IR transmission, in the region above 10 kHz are prensent even if there is no light impingin on the photodiode. They are likely be due to electronics.
1. The installment of camera for IR reflection
Since we have received the filter to attenuate green beam, we tried to install the camera for IR reflection today. Since the power of IR is pretty high, we also used an OD filter(factor of 2) and a partly-reflected mirror to reduce the power. During the adjustment, we can see on the screen that there are two points. The small one is on the right and below side of the large one. Then we tried to change the angle of that partly-reflected mirror and accordingly the camera. At a certain point, we could make it a round point. We believed that that was a good angle.
Then we locked the cavity, this reflection of one point changes to two points.(See attached video 1) We thought that we were looking at the TEM10 mode. However, it should be the superposition of TEM00 and TEM01. According to our knowledge, this reflection can tell us the information of alignment. That means if it is the right reflection, we can use it to refine the cavity. Since we have already been around the best position, so we move the Input mirror around this position to check. What we could see is only these two points changed brighter and weaker one after another.(See attached video 2) Even when we tried to misalign the pitch of End mirror, we couldn't see the TEM01 mode on the screen.(See attached video 3)
Video Note: the interesting part of video are all around the end of video. You can find them by using the date 20180226. You can check them through this link. https://drive.google.com/drive/folders/1v7oSk0d6ONPN-NZTNcYjIuG0ip8XCZOn?usp=sharing
After checking all the things listed above, I felt gradually that it is not the right reflection. But we cannot figure out why we can see it. We need an expert and tomorrow Raffaele will come^_^
2. Attempt to acquire IR error signal
Due to the FWHM of IR is much more narrow than GREEN. The locking accuracy of IR needs to be evaluated by having the error signal directly. Owe to the EOM modulation for SHG, we have the modulation for infrared luckily. So we separate the output of its signal generator by using SMC-type connector(to avoid signal reflection see attached picture 1). We use the reflection from the cavity as RF signal.
After separation, we succeeded in recovering the error signal for the SHG. By the way, we made this error signal pk-pk value larger than before.(See attached picture 2 and 3) Also we took the picture of different modes we got for SHG.(See attached picture 4)
For infrared error signal, we make the AOM modulate. That means the AOM driving frequency change around the best point with a certain frequency(It's like the ramp signal while we looking for the error signal). But we cannot find the error signal for IR.
For the fail of IR error signal, we found the bandwidth of IR reflection PD is not high enough. Now it is PDA36A(350-1100nm, 10MHz BW, 13mm**2), but our modulation is 15MHz. Certainly we cannot demodulate it. Besides, we also need to check if the demodulation board can work well tomorrow.
I characterized the absorption of the central part of the tama-sized sapphire substrate sample#2
Pump power 10W (after the chopper). Max laser current: 7.5A
After aligning the system with the surface reference sample, and scanning the bulk reference sample for calibration (with power=30mW), I made a scan of the sapphire substrate.
Changing from the 3.6mm thick reference sample to the 60mm thick sapphire substrate we have to move backward the imaging unit by 24.9mm. Then recenter the probe on the PD maximizing the DC.
From the scan plot we can recognize the 2 surfaces of the sample (when the absorption signal drops to 0) and associate the translation stage Z coordinate to the sample coordinates. Indeed the apparent depth of the sample is different from the real one as explained in entry 242 , Since the sample is 60mm thick, it is convenient to define the sample Z coordinate to be 0 at the first surface and 60 at the second surface. On the translation stage reference system the first surface is at 44mm and the second surface is at 77.5mm. In other words, during a scan, the crossing point pump-probe travels in the sample about 1.8 times faster than the translation stage that mpoves the sample.
I made circular maps on the XY plane at several depths in the sample. Then 2 rectangular maps on the XZ and YZ planes. A 3d overview can summarize all the maps together in comparison with the sample size (drawn as 2 circles for the surfaces boundaries)
The circular maps resolution is 100um x 100um
The rectangular maps resolution is 100um x 1mm
the waiting time from one point to the next one is 1s, according to the average and median filters of order 10.
the color scale is up to 200ppm/cm to make the maps more readable and cover most of the color range. However, on the second surface of the sample, there are some absorption peaks up to 1400 ppm/cm which are probably due to some dirt on the surface or some polishing defects
I made a mistake in entry672. The noise is injected in the position of CAVITY(G3). So we need to use open loop TF and G1,G2,G3 to calibrate this signal.
So I measured the open loop transfer function, it agrees with the simulation of Eleonora's thesis. For the measurement, the magnitude for 20Hz is about 10^4.(See picture 1)
Then using CH2*(open loop/(G1*G2*G3))=CH1. From this formula, we can do the calibration like entey672 again. This time, the result of calibration approximatly equals to the measurment.
By the way, I also tried to insert noise of 500Hz. It aggres with the prediction, the suspension can filter this high frequency noise. After injecting this noise, we cannot see any peak on the spectrum of PZT or error signal.(See comparation of picture 2 and 3)
Mechanical transfer functions of local controls as of 21 February 2018.
Noise amplitude: 500mV (white noise injected into port NOISE 2)
Here is attached noise spectra of the filter cavity transmission and reflection. As for transmission spectra, data measured in two different frequency span were put together afterwards to retain good resolution in lower frequency range. There are spikes on IR transmission in the range above 10kHz which do not exist in reflection nor in green. We also examined IR beam before entering the cavity and confirmed that there were not such spikes (data not shown here).
We have verified that the spikes oberved in the IR transmission, in the region above 10 kHz are prensent even if there is no light impingin on the photodiode. They are likely be due to electronics.
This calibration factor is measured while sine wave noise frequency is 28kHz. I also did the measurement for different sine wave frequency. It is shown in picture 2. For this, I have a question. Is this frequency dependence related to the transfer function of our control servo?
From the picture 2, we can read the peak value of this noise. It is 45.7407uVrms. And according to the block flow, we can calibrate this value to error signal. The formula is Err_V = K(V/Hz) * S_Hz/(1- (f/f_0)^2).
K is 3.1e-3, S_Hz is V_RMS (V) * 100 * sqrt(2) * 2e6 Hz/V=12937Hz, f_0=1.45kHz. In this case Err_V=38.8Vrms. However we read from the spectrum, it is 1.6mVrms.
We also checked the situation if there is not noise injecting to the end mirror. It is shown in picture 4 and 5. By comparing them, I found the 20Hz disappear. But the noise level for the whole frequency band(from 1Hz to 50Hz) has decreased.
Conclusion:
1. The shaking of EM can cause precisely the same frequency noise in PZT. And also the harmonic peak.
2. In the aspect of error signal, the shaking of EM noise will be distributed from one frequency to the whole band. This distribution may come from filter.
3. By using the transfer function we created before, we cannot predict the behavior of error signal in the low-frequency band.
Next step:
To solve the problem of transfer function for low-frequency.
We measured the misalignment again by changing the driving frequency of AOM. The measurement shows it is around 0.168. And the standard deviation is 0.0165.(See attached picture 5) So the residual misalignment is 17% (+/-2%) .
Besides, we injected a sine-wave noise into the local control of optical lever. We choose the frequency of 20Hz, 200Hz and 500Hz. The amplitude is 1V. The noise is injected into EM (yaw and pitch). The 20Hz is appreciable on the transmission of green.(See attached picture 1)We are sure the noise is injecting to EM.(We can see it from the local control like picture2) And the frequency of the noise we are injecting is the same with the frequency we can see in the transmission signal of green light. At the same time, we measured the noise spectrum of error signal. We found it has no difference from the case without noise injecting.(For all the test, they have no difference. See picture 3 and 4)
This perturbation test may tell us that the angular to length coupling is low, so that its affect for error signal is below the present noise level.
I use the data to calculate the standard deviation of IR transmission and GREEN transmission. For green, it is 0.0286. For IR, it is 0.1235.
The result is also shown as attached figure. Next step is to figure out the reason of instability.
Participants: Eleonora, Matteo L., Raffaele, Tomura
In the past days we kept working on the characterization of the IR lock, with the main purpose of measuring the cavity losses and the decay time.
ROUND TRIP LOSSES
We observed that in good alignement condition (IR trasmission above 1.5 V), the fluctuations of the transmitted and reflected power were much less than what observed before.
In this condition we were able to measure a change in the cavity reflectivity when the cavity is resonant and when it is not and give a preliminar estimation of the round trip losses (RTL)
In Fig.1 the trasmission and the reflection of the IR are shown when a set of lock/unlocks of the cavity was done. The reflected light has been focused on a photodiode using a lens with f = 50 mm. Et the beginning of the measurement the IR light has been blocked to measure possible offsets of the photodiodes.
The technique used to switch from resoant to not resoant state was to suddenly change the driving frequency of the AOM of 5 kHz. By using the values of the reflected power in the two states (resonant and not resonant) as explained in detail in the attached pdf we estimated the RTL to be about 80 +/-12 ppm, corresponding to 0.26 ppm/m. The error is mainly do to the residual fluctuations of the refected power when the cavity is locked. The associated squeezing degradation is reported in FIg 2.
The presence of light not coupled in the cavity (mismatching/misalignement) normaly reduces the measured losses and has to be compensated in order to have a real estimation of them. In the previous calculation I assumed a mismatching of 15%.
[An idea of the impact of the mimastch compensation: assuming no mismatching the computed losses are 70 ppm while with 20% of mismatching they becomes 85 ppm. (Details about this can be found in my thesis at pag.101)]
According to the simulation we expected about 55 ppm of RTL (40 ppm from flatness, 10 ppm from rougness/point defects and 5 ppm from trasmission and absorpition). Note that losses from small angles scattering (between mrad and few degrees) have not been considered in this loss budget.
DECAY TIME and FINESSE
A preliminar estimation of the decay time has also been done. To do that we used different tecniques: bringing the cavity suddenly out of resonance (by stopping the lock with the servo or changing the AOM driving frequency) or cutting the light in input. The transmission and reflection in this 3 cases are reported in the second pdf attached)
A fit of the transmitted power for the first measurement shows a decay time of 0.0027 s, corresponding to a finesse of 4250 (Finesse = pi*FSR*decaytime ). See third figure attached.
Assuming the nominal reflectivity of the mirrors, this value is compatible with RTL of about 100 ppm.
A better analysis of the decay time, with an estimation of the error bars will be done soon.
Participants: Eleonora, Tomura
We conducted characterization of two auxiliary lasers and faraday isolators.
For two aux. lasers, beam dimension evolution along propagation were measured. Laser power during measurements were approx. 260 mW, which corresponds to 1.2A in laser current. Attached figures shows bean waist sizes and position. The origins of x-axes were set at aperture of laser housing. The waist positions were somewhat different from what expected.
The beam polarizaiton purity was confirmed with half/quarter waveplate. It was 96.7% and 99.9% for aux1 and aux2, respectively.
Beam transmission through FI was also measured using aux laser as a light source. Pictures of two FIs were attached.
For first FI (thorlabs), transmitted power was 213 mW out of 265 mW, which means 80.4%.
For second FI (Gsaenger), it was 79%.
Great!
Participants: Eleonora, Raffaele, Matteo L.
We have installed in the end bench a thorlab passband filter (FL1064-10) on the path of the IR beam after the harmonic beam splitter. This allowed us to get rid of the residual green light and finally to monitor the beaviour of the IR light when the cavity is locked.
By changing the driving frequency of the AOM installed on the green path we were able to induce a frequency shift between the the green and the IR frequency in order to have both resonant at the same time.
NOTE THAT: since the AOM is put on the green path, the change in the frequency which it induces is compensated by the servo with a change on the IR which is half of the frequency change in the AOM. This means that a shift of 1 MHz in the driving frequency of the AOM corresponds to a shift of 500 kHz in the frequency of the IR light.
In the following are reported some preliminary results that we were already able to obtain:
TEM00 resonance and FSR: The resonance frequency of the TEM00 has been found at about 109.366 MHz. (The standard driving frequency of the AOM is 110 MHz). We have verified that it occurs every time we shift the AOM frequency of 1 MHz, meaning that the FSR is 500 kHz, as expected.
Cavity finesse for IR: We did a rough estimatin of the IR linewidth by slightly changing the AOM frequency in order to scan the IR resonance. We drove the AOM in order to find the maximum in the transmission (about 1.5 V) and then we checked the frequency shift necessary to get half of the maximum. Repeating this procedure few times, we found a FWHM of about 110 Hz +/- 10. This corresponds to a finesse of about 4500 +/- 450, which is consistent with what we expect. The big error is due to a large fluctuation in the trasmitted power.
HOM resonance frequency: The first order modes have been found at a AOM frequency of 108.970 MHz (and FSR multiples). It means that the the difference in the resonance frequency betwen the fundamental and the first HOM is about 198 +/-1 kHz. This value is in very good agreement with what expected, considering the cavity g-factors computed by using the RoC values measured at LMA.
According to LMA measurement, we have R_in = 436.7 m and R_end = 445.1 m which correspond to a frequency shift given by
δν = (FSR/pi)* acos (sqrt (1-L/R_in)*(1-L/R_end) = 198.1 kHz
We also observed that different modes of the first order resonate at slightly different frequencies, which is likely to be an effect due to the astigmatism. More investigations have to be done.
Residual misalignment: by comparing the trasmitted power of the funtamental mode with that of the first HOM we estimated the misaligment to be between 20% and 25%.
Lock accuracy: Since we don't have an error signal for the IR lock, it is not easy to evaluate the lock accuracy. Anyway the fact that resoanance can be scanned by shifting the AOM frequency gives us a lower limit that is surely better than few tens of Hz, that is much better than the accuracy on the green lock, estimated to be about 100 Hz. This is compatible with the hypothesis proposed by Matteo B, according to which the IR pole (at much lower frequency than the green one) contributes to filter the high frequency laser noise, improving the lock accuracy.
Stability: Once put on resonace the IR seems to be stable as long as the lock is mantained, anyway we observed that when the cavity unlocks and relocks it is not always possible to get the IR resonant again. This is probably due to the fact that only half of the green locking points are also good locking points for the IR.
The attached picture shows the transmitted beam at resonance for the IR (top screen) and green (bottom screen)
Great!
Participants: Eleonora, Matteo L.
Last Thurday, while measuring the hight of some optical component on the bench, the green beam got a bit misaligned. When we realigned it, we observed that the trasmission when the cavity is locked is higher than usualy (1.5 V). The beam seemed still centered on the cavity mirrors and the IR is still well aligned.
We observed that the references of the PR and in particualr BS have slightly changed in yaw and we marked a new one.
At some point we had a problem of spikes with PR yaw error signal. We observed that this was the only one very far from zero. The spikes disappear as soon as we put it backto zero.
Participants: Tomura, Eleonora
As reported in entry 654, we had to move the yaw of SM2 dicroic mirror in PR chamber to avoid the IR reflection to touch the viewport side. Since after the last 1000 steps done, we were not able to get a good IR alignent anymore, yesterday we moved back SM2, hoping to find a good postion where we could have both a good alignement and the maximum of the power in reflection.
The input power is about 16 mW.
The situation is such that in the original position (before last thurday) the aligment can be very good but the beam hits the viewport side and the reflected power is very low (less than 1 mW). In order to have the maximum reflected power possible (about 12.5 mW) SM2 has to be moved so much and it is not possibile to recover a good alignent. (We remarked that in this case the cavity seems very mismatched, but we where not able to improve the situation by moving the last IR lens on the bench).
Today we tried to put back SM2 towards the original position and to fine-tune its position in order to have the best compromise between good alignement and high reflected power.
We did the followoing series of steps (velocity 500)
-300, -300, -300, -1000, -1000, -1000, +1000, +1000, +500, +300, -300, -300, +200, -100
In the end, we stopped in a position where we can have quite a good alignement (but not the best possible) and about 11 mW of reflected power. By looking with the camera it is not evident that the reflection is touching the viewport side.
We also moved the last IR lens back, from 20 mm to 17 mm.
In the attached files we recoreded the new reference for the green beam: the white cross corresponds to the orginal position and the black point corresponds to the maximum displacement done by SM2.
Considering the maximum reflection we could get (12.5 mW over 16 mW), we estimated the injection losses in vacuum to be about 22% (11% single pass) which is not much higher than we expected.
We measured the beam profiles of all the 3 lasers we have. See the plot
We aligned them at the cross point and maximized the signal using the surface reference sample.
Then we made a scan of the surface and the bulk reference samples.
Yesterday I did a very preliminary check of the AOM operation. After locking the filter cavity I have changed the driving frequency of the AOM (which is normaly set at 110 MHZ) and I have looked at the correction sent from the servo to the laser piezo.
In the attached picture you can see in yellow the trasmission of the filter cavity and in blue the laser correction (from PZT mon).
I changed the AOM driving frequency in steps of 100 kHz each time. Most of the time this change causes a very short unlock of the cavity, as it can be seen from the picture. Nothing can be noticed in the correction signal.
Since the piezo calibration is 2 MHz/V and the channel PZTmon has an attenuation of 100, a change of 100 kHz should result in a change of 2 mV in the correction, which is likely not visible. I tried to do bigger steps but the cavity loses easly the lock.
I observed that the correction is about a factor 3 bigger than that measured this summer. (See last picture of entry 475)
Participants: Raffaele, Eleonora
As reported in entry 649, while improving the IR aligment we realized that the power of the reflected beam was very small. After some investigations we were able to see with the IR camera that the reflected beam was hitting a side of the inner part of the viewport. in order to avoid this we decide to slightly change the positon of the diachroic mirror SM2 (where green and IR recombine) in order to change the way the beam passes through the faraday.
We moved the yaw picomotor of SM2 several times (velocity 500, direction +), trying to recover the IR aligment with the two steering mirrors on the bench after each picomotor movement.
We did the following series of steps:
10, 10, 100, 300, 300, 600, 600 (Thursday) and 600,1000 (Friday)
Before the last movement (1000 steps) we were able to have a quite good alignment and a reflection between 11 and 12 mW (the input was about 16 mW), nevertheless we were still able to see some IR scattered light inside the viewport. After the last movement (1000 steps) the scatterd light was no more there but we were not able to recover a good aligment.
Sometimes, while recovering the aligment we had the impression that the matching was worsened (more LG modes), so we have changed the postion of the last IR lens different times in a range from 7 mm (original postion) to 22 mm.
Note that since we moved SM2, one of the two green references (the reflection from SM2) has changed accordingly. See attached picture.
Participants: Manuel, Eleonora, Tomura
Today, after asking Takahashi-san, we removed most of the optics intalled on the optical table placed at the bottom of the stairs in TAMA central building. There were three lasers ( two 532 and one 1064) and many other optics which have been temporarily stored in a plastic box below the table. Currenty the table is being used to test the two auxiliary lasers and faraday isolators for the filter cavity experiment.
I attach some pictures taken to the table before we removed the optics.
